We are only days away from the release of GPT5 and it looks like it's going to be a game changer in some positive and possible many negative ways.
GPT4 was dumb in many ways but also extremely "intelligent" as an extension to your intelligence, meaning that it did know what you didn't and with the right prompt could understand what you were after. That and only that: Perfect.
GPT5 is going to be very different, with many add-ons which may completely change the experience. Most people will see this as "progress" at least at the beginning but I am not sure at all!
The ability to take more "context" into consideration will make it closer to AGI. Good. Although, now it's going to take the past into consideration. How dangerous can this be? Could it be the beginning of the really never forgetting internet, in which case we truly do have a problem? Whatever you have done, wherever you have been, whoever you have met will now catch up with you retrospectively. No problem? Think again: Now thanks to AI we can make inferences. The machine will quickly "know" you better than you do. This is the key to "pre-crime" or more insidiously manipulation on a massive scale, individually.
With AI, any picture can retroactively be geo-localised. Persons identified. Past mails, letters, videos, pictures analyzed and understood in their context. How long before the AI creates a perfect "image" of you through time? What can be done with that?
And that's just the past. What about the present? How long before you are recognized instantly on EVERY camera linked to AI? How much privacy will you have left if almost anybody can know where you are at any time? Remember Jason Bourne 20 years ago? The CIA using advanced computers could never quite figure where he was in real time. Now the AI will know where you are before you arrive where you want to be! If you are not yet frightened, you are not paying attention.
In the past, laws have always been voted with "good" or at least valid intentions in mind and soon after used for a completely different purpose. Here we'll get the problem on steroid. New laws about hate crime and misinformation in Europe and the US are already quite dangerous in themselves but with AI extensions, any malevolent government will soon be in a position to do things that the worst dictators could not have dreamed about. How much freedom can there be in such a world? The more you think about it, the more the danger of the technology becomes obvious. We're on our way to giving birth to a monster!
No comments:
Post a Comment